Measurement first.
Enhanced Conversions, Meta Pixel and CAPI, a weekly Salesforce import, one UTM taxonomy, and multi-touch attribution. The account could finally see pipeline, not form fills.
Case study · Legal AI
The Harvey teardown.
One working channel, $122,500 a month, and no reliable way to prove what the rest of the budget was doing. Here is what changed in 90 days.
Harvey · Legal AI · Google, LinkedIn + 4 new channels
A $122K/month program with one working channel, rebuilt into a six-channel pipeline engine.
Harvey was spending $122,500 a month across Google and LinkedIn, and only branded search was pulling its weight. Non-branded Google ran at $600+ per acquisition with nothing wired back to Salesforce to prove otherwise. The teardown started with a hard look at what the money was actually buying.
Before
$20M
paid-attributed ARR. Two channels, one of them working.
After 90 days
$76M
paid-attributed ARR. Six channels, fully instrumented.
+$56M
added paid-attributed ARR
3.8×
revenue versus the starting state
2 → 6
channels live and measured
90 days
from audit kickoff to full program
The diagnosis
Technically active. Not actually a demand engine.
60%
of spend sat in non-branded Google at $600+ CPA, on Quality Scores of 3 to 4. The biggest line was the weakest one.
6 of 9
channels were untested. Meta, syndication, Reddit and display had never been switched on.
0
feedback loop to Salesforce. No Enhanced Conversions, no Meta Pixel, and a 60 to 180 day sales cycle read through last-click. Every channel looked worse than it was.
1 layer
of LinkedIn targeting, job title or company size but not both, and zero retargeting despite the Insight Tag already firing.
“The ad account is technically active, but the paid media program is not functioning as a demand generation engine.”
The rebuild
Fix the measurement, then earn the right to scale.
Nothing scaled until the account could tell the truth. Measurement went in first, then the budget moved to whatever the data actually rewarded.
01
02
Rebuilt the spend.
Non-branded Google was paused and rebuilt around intent, with RLSA and competitor conquesting added. LinkedIn retargeting launched first, the fastest high-ROI unlock in the account.
03
Opened new channels.
Meta, then content syndication in the legal press, then Reddit and programmatic. Each one aimed at three sharply defined buyers, from Am Law 200 partners to in-house GCs.
Oct 2025
Measurement and infrastructure
UTM taxonomy shipped, Enhanced Conversions verified in GA4, Salesforce import tested, Meta CAPI firing. Non-branded paused and restructured.
Nov 2025
Google, LinkedIn, Meta live
Non-branded and RLSA live on weekly ROAS. LinkedIn retargeting and Thought Leader Ads from three Harvey team members. Meta and first syndication placements benchmarked.
Dec 2025
Optimize and expand
Competitor conquesting on Google and LinkedIn, multi-touch attribution live in Looker, demo no-show recovery running. Phase 2 approved on real CPL and pipeline data.
The outcome: six of nine channels online in 90 days, five of them at or under target CPL, and paid media that finally reported in pipeline instead of clicks.
Start here
Want results like these on your account?
Book a 30-minute teardown. We'll show you the three things costing you the most right now, on your numbers. You keep the findings either way.
30 minutes. Direct with Jer.